Objective cognitive impairment is a feature of Lewy body dementia (LBD), and computerised attentional tasks are commonly used as outcome measures in interventional trials. However, the reliability of these measures, in the absence of interventions, are unknown. This study examined the reliability of these attentional measures at short-term and longer-term follow-up stages. LBD patients (n = 36) completed computerised attentional tasks (Simple and Choice Reaction Time, and Digit Vigilance (SRT, CRT, DV)) at short-term (Day 0 – Day 5) and longer-term (4 and 12 weeks) follow-up. Intra-class correlations (ICCs) were calculated to assess test-retest reliability. At short-term, the reciprocal SRT, CRT and DV mean reaction time to correct answers, the reciprocal DV coefficient of variation, and reciprocal power of attention (PoA) all showed excellent levels of reliability (all ICCs > 0.90). The reciprocal PoA showed the highest level of reliability (ICC = 0.978). At longer-term follow-up, only the reciprocal PoA had excellent levels of reliability (ICC = 0.927). Reciprocal SRT, CRT and DV reaction time to correct answers, and the CRT coefficient of variation values, showed good levels of test-retest reliability (ICCs ≥ 0.85). Contrary to expectations, most attentional measures demonstrated high levels of test-retest reliability at both short-term and longer-term follow-up time points. The reciprocal PoA composite measure demonstrated excellent levels of test-retest reliability, both in the short-term and long-term. This indicates that objective attentional tasks are suitable outcome measures in LBD studies and that the composite PoA measure may offer the highest levels of reliability.